The Weekly Edge: GraphRAG with Neptune, Kuzu, & Neo4j [25 July 2025]

Welcome back to the Weekly Edge! 

This new blog series from gdotv curates the tl;dr of graph tech news and resources from the past week (or so). 

This week’s edition features two books (one of them free), GraphRAG resources for multiple graph databases, a great graph podcast series, a couple reads related to graph query languages, and a little something from us at gdotv. Let’s dive in!

Podcast: Graph Power Hour Podcast with Paco Nathan – GraphRAG at Scale with Amazon Neptune & Amazon Bedrock

If you’re reading this and you don’t know who Paco Nathan is, then let me introduce you to the Gandalf of Graph Databases. (Catch my interview with Paco Nathan on AI realism and knowledge graphs here.) 

The Graph Power Hour podcast hosted by Paco Nathan – the Principal DevRel Engineer at Senzing – is an essential resource for graph practitioners who want to keep up with developments in the space and encounter new ways of thinking with graphs. 

In this episode of Graph Power Hour, Paco interviews Ozan Eken and Nicole Moldovan (both at AWS) about the new GraphRAG capabilities working with Amazon Neptune graph database and Amazon Bedrock (foundational models for genAI). It’s well worth a watch/listen. 

Free Book: Essential GraphRAG by Tomaž Bratanic & Oskar Hane

To borrow from Mugatu, that GraphRAG is so hot right now. The good news is you can catch up with the essentials of using knowledge graph-enhanced Retrieval Augmented Generation (RAG) in this book Essential GraphRAG by my former colleagues Tomaž Bratanic & Oskar Hane. 

With a foreword written by…Paco Nathan(!), this book takes you through GraphRAG from top to bottom with examples you can code and run yourself using Neo4j and Cypher. This book from Manning Publications is being made available for free for a limited time via sponsorship from Neo4j, so grab your copy today. (Thanks to Michael Hunger for sharing!)

Blog Post: Improving Text2Cypher for Graph RAG via Schema Pruning

Going three for three on GraphRAG, here’s another great resource on using (and improving) Text2Cypher for GraphRAG. This blog post by Prashanth Rao shows you how schema pruning – removing unnecessary parts of the graph schema and keeping only what’s relevant – dramatically improves Cypher generation from natural language using Text2Cypher. Prashanth uses the LDBC-1 dataset, Kuzu, and BAML in his examples to frame schema pruning as a context engineering problem – with a context engineering solution. 

If you’re interested in learning more about this GraphRAG use case, Prashath and Dennis Irorere (from GraphGeeks) presented on “Agentic Workflows for Graph RAG: Building Production-Ready Knowledge Graphs” at the virtual OSDC Agentic AI Summit on July 24th using Kuzu and G.V() in a live demo. (Recordings of the session are available on demand for all summit ticket holders.)

Book: Getting Started with the Graph Query Language (GQL)

Not much of a brag, but I remember working in graph databases before the ISO standardization of Graph Query Language (or GQL) *pushes glasses up my nose.* But these days, kids have it easy with the release of this new book Getting Started with Graph Query Language (GQL).

So hop on the GQL bandwagon with your copy of Getting Started with Graph Query Language (GQL) from Packt Publishing and written by the Ultipa graph database team: Ricky Sun, Jason Zhang, and Yuri Simione. 

Graph Query Language is the SQL for graphs and was influenced by Cypher, GSQL, and SQL itself. While not every graph database vendor supports GQL at the moment, it’s the standard that the industry will undoubtedly be headed toward. (For what it’s worth, G.V() offers GQL query editor support for every graph database that uses it!) 

Release: G.V() 3.34.79 with Oracle Graph Support, New No-Code Data Exploration & More

I promise to not always talk about G.V() – that’s pronounced “gee-dot-vee” – in the Weekly Edge, but hey, we still gotta pay the bills like anybody else. This past week, we released G.V() 3.34.79 and it’s got a juicy package of new features and integrations.

Here’s the tl;dr:

• Oracle Graph support for Oracle 23ai using the SQL:2023 query language
• New no-code complex path pattern searches
• Single-file database storage with Kuzu graph databases
• Upgraded syntax handling for Cypher and Gremlin query languages
• Single sign-on and user federation for G.V() Web
• A few other miscellany

If you’re looking for a graph database client and IDE – or you just need some seriously easy graph visualization – why not give G.V() a try? 

***

That’s it for this week’s picks. Got something you want to nominate for inclusion in a future edition of the Weekly Edge? Ping us on on X | Bluesky | LinkedIn or email weeklyedge@gdotv.com.